Supporting Understanding Through Task and Browser Design

Supporting Understanding through Task and Browser Design Jennifer Wiley (jwiley@uic.edu) Department of Psychology, 1007 W. Harrison Street (M/C 285) Chicago, IL 60607 USA Abstract While electronic text offers the potential to explain, illustrate, and scaffold understanding in powerful new ways, few studies on educational use of electronic text resources have shown significant learning gains, or even measured learning outcomes in controlled experiments (Chen & Rada, 1996; Dillon & Gabbard, 1998). In a follow-up to previous studies (Wiley & Voss, 1999), the present experiments study the effects of different tasks and browser designs on navigation and reading patterns, as well as on memory and comprehension measures from electronic text. These studies have revealed that only when both the task and the design support integration (such as in a two-windowed browser) and students are explicitly directed how to use the feature, do students take advantage of the flexibility of the multiple-source environment, integrate across sources, and achieve the best level of understanding. Introduction One promise of using electronic text in the classroom is the potential for students to search for, access and read multiple forms of information about a topic. Since the search for and navigation of digital documents is student-initiated, requires student interaction, captures the student’s interest, proceeds at the student’s own pace, and allows for flexible navigation and juxtaposition of multiple sources, a number of theorists have suggested that the web might be a powerful tool for student instruction (Beeman, et al 1987, Spiro & Jehng, 1990). This optimism is consistent with a number of recent cognitive studies demonstrating that activities that require readers to engage in active, constructive and integrative tasks lead to the best understanding of the subject matter (e.g., Chi, de Leeuw, Chiu & LaVancher, 1994; McNamara, et al., However, a review of the literature on educational use of electronic text yields two striking conclusions. First, students generally fail to utilize hypertext links and multiple window capabilities effectively, if at all, as they read (Foss, 1989; Gordon, et al., 1988). This is especially true of novice users (Foltz, 1996; Gray & Shasha, 1989; Tombaugh, Lickorish & White, 1987). And second, few studies on educational use of electronic documents, whether from stand-alone hypermedia or the World Wide Web, have actually shown significant learning gains (Chen & Rada, 1996; Dillon & Gabbard, 1998). What is critically needed is for experiments to determine which specific instructional contexts may allow for effective educational use of electronic text. Although there has been a great deal of evaluation on effective browser design from a Human-Computer Interaction (HCI) standpoint, effectiveness has been measured largely in terms of efficiency of search or ease in information finding (Dillon & Gabbard, 1998). While such fluency measures may be related to some extent to the amount of information a person is able to recall after reading from computer screens, they may not be correlated with whether a person develops an understanding of the text that is being read. A number of studies have shown that conditions that produce the best surface memory for text are not the best conditions for producing the best understanding of text (e.g. McNamara, et al., 1996; Mayer, 1999; Wiley & Voss, 1999). While surface memory may correlate with the fluency or ease of information processing, understanding may depend to some extent on the need to put effort into developing an underlying representation or situation model of the text (Kintsch, 1998). Thus, previous assessments from an HCI perspective cannot reliably indicate which screen layouts will be most effective for promoting understanding from electronic text. Given this educational goal, browser design must be evaluated specifically using measures of conceptual learning. In a review of the published studies on hypermedia and learning outcomes between 1990 and 1996, Dillon and Gabbard found only 11 studies that performed controlled experiments on hypermedia and learner performance. Of those 11 studies, there were only four results that actually seemed to suggest an advantage for learning from hypermedia over paper. In the majority of studies there was no clear difference between learning from hypermedia and learning in a control (more traditional) setting. While this may be viewed more optimistically as evidence that learning from electronic resources may sometimes be no worse than traditional classroom methods, there is hardly an overwhelming body of evidence that the web can generally be relied upon to provide an enriching instructional experience. Of the four studies that netted positive learner performance, only two seem to indicate that hypertext may allow students to engage in learning at a more conceptual level. One of these reported that students learned to recognize aircraft more efficiently and effectively when they were able to view the images